In the long term evolution (LTE) system, reliability and performance of the transmission are greatly increased by using the hybrid automatic repeat request (HARQ) transmission technology. For frequency division duplex (FDD), the uplink subframe and downlink subframe are symmetrical, so it is quite simple to perform the HARQ process, and the timing correspondences are also quite consistent. For time division duplex (TDD), the number of uplink subframes and the number of downlink subframes vary greatly in different configurations due to the difference in the resource assignment ratio of the uplink subframe and downlink subframe. Therefore, the HARQ feedback in TDD is more complicated, and the HARQ feedback has two processing methods: HARQ bundling and HARQ multiplexing. These two methods compensate each other's defects and may be configured through the network side.
In the case of better channel conditions, the HARQ bundling method may greatly reduce the number of bits used in feedback, but also has a drawback. When a packet is wrong, multiple packets associated with the packet need to be retransmitted, and in the case of poor channel quality, the overhead for retransmission is relatively high. How to feed back a HARQ when the detection of the downlink is missed is a significant consideration in the third generation partnership project (3GPP) protocol, while no regulations are given for a specific case in which the detection of the uplink grant is missed.
If the uplink grant of DCI0 is received on the n-km subframe, the terminal needs to feed back a HARQ acknowledgement (ACK)/negative-acknowledgement (NACK) using a physical uplink shared channel (PUSCH) on the uplink subframe n.
If the uplink grant of DCI0 is not received on the n-km subframe, the terminal needs to feed back the HARQ ACK/NACK using a physical uplink control channel (PUCCH) on the uplink subframe n. If the uplink grant of DCI0 is transmitted by the network side on the n-km subframe and the terminal misses the detection of the uplink grant of DCI0 for some reason (missing detection may be caused by many reasons, some of which are unavoidable, such as a sharp deterioration of the channel condition), the terminal mistakenly considers that the uplink grant of DCI0 is not transmitted by the network side, while the network side considers that it has transmitted the uplink grant of DCI0. Then the network side expects to receive the HARQ ACK/NACK on the PUSCH. However, the terminal transmits the HARQ ACK/NACK on the PUCCH, causing that the network side misses the uplink HARQ ACK/NACK, and leading to following consequences.
(1) If a UE intends to feed back the HARQ ACK and the network does not receive the feedback, the network will consider that the UE did not receive any data, then all data packets are retransmitted, resulting in a retransmission in vain and reducing performance.
(2) If the UE intends to feed back the HARQ NACK and the network does not receive the feedback, the network will consider that the UE did not receive any data, then all the data packets are retransmitted using a NDI flipping method. When the UE receives the NDI flipping, the UE empties the HARQ buffer and deleting the received data in the HARQ buffer, so the transmission needs to be repeated from the beginning, losing an advantage of the HARQ in incremental redundancy decoding.
If a DCI format 0/4 is lost, the UE does not know whether the eNodeB has transmitted the DCI format 0/4 since the UE does not receive the DCI format 0/4 at all. The UE feeds back the ACK/NACK according to currently received information, and the UE feeds back the ACK/NACK on the PUCCH. In addition, the eNodeB does not receive the ACK/NACK on the expected PUSCH. This is because the UE does not receive the DCI format 0/4 and the resource for the ACK/NACK feedback is changed (changed from the PUSCH to the PUCCH), instead of because the downlink transmission is not successful. Since the protocol does not specify this special case, the eNodeB detects no uplink information on the PUSCH and considers that the mobile phone did not receive the downlink data, so the eNodeB retransmits all the data that has been successfully transmitted, resulting in poor efficiency.